AUCFitted

class rectools.metrics.auc.AUCFitted(outer_merged_enriched: DataFrame, n_pos: Series, n_fp_insufficient: Series)[source]

Bases: object

Container with meta data got from _AUCMetric.fit method.

Parameters

outer_merged_enriched (pd.DataFrame) – Recommendations outer merged with test interactions. Table has Columns.User, Columns.Item, Columns.Rank. Precomputed columns include “__test_positive”, “__tp”, “__fp”, “__fp_cumsum”, “__test_pos_cumcum”. All ranks for all users are present with no skipping. Null ranks are specified for test interactions that were not predicted in recommendations.
n_pos (pd.Series) – Number of positive items for each user in test insteractions.
n_fp_insufficient (pd.Series) – Number of false positive items for each user in outer_merged_enriched that had at least one false negative. This users will be checked for insufficient cases processing.

Inherited-members

Attributes